You are on page 1of 7

RESEARCH ARTICLE

Diet-Cancer Related Beliefs, Knowledge, Norms,


and Their Relationship to Healthful Diets

RUTH E. PATTERSON,! ALAN R. KRISTAL,! JAMES C. LYNCH,I,2 AND EMILY WHITE 1


ICancer Prevention Research Program, Fred Hutchinson Cancer Research Center, Seattle, Washington 98104;
and 2Department of Biostatistics, University of Washington, Seattle, Washington

ABSTRACT Our objective was to examine the prevalence a convergence of results from laboratory and epidemiologic
of diet- and cancer-related psychosocIal constructs in a popu- research that suggests that fat can increase the risk of certain
lation-based sample and their association with healthful diets. cancers while fruits, vegetables, and whole grains are pro-
We administered a random digit dial survey on cancer-related tective. 2- s There are many ongoing activities to deliver the
risk behavior to 1972 adult Washington state residents. The
diet and cancer message to the public, including govern-
psychosocial constructs were belief in an association between
ment health promotion programs, food advertising, and
diet and cancer, knowledge of National Cancer Institute diet
community intervention trials .H2 However, there is still
recommendations, knowledge of fat and fiber composition of
foods, and perceived pressure to eat a healthful diet (norms). much to learn about the most effective means of improving
Diet measures were self-reported healthful diet changes over diet habits in the general public. In particular, there is little
the previous 5 years, percent energy from fat, and dietary fiber. consensus on the components of community-based inter-
Age and education were significantly (p ::; .001) related to the ventions that are most effective for initiating and maintain-
constructs. Older adults (60+) had the lowest levels of belief, ing dietary change.!3
knowledge, and perceived norms. Participants with strong Programs to improve eating patterns are more likely to
beliefs in a diet-cancer connection consumed 1.1 grams more be effective when they are based on established theory and
fiber than those with no belief (p for trend ::; .001), and research on changing health-related behavior. 14 Of particu-
participants with excellent knowledge of food composition lar interest for designing mterventions to promote healthful
consumed 2.1% less energy from fat than those with low
diets are constructs derived from social cognitive theory,
knowledge (p for trend::; .001). Perceived norms were only
the theory of reasoned action, consumer information proc-
weakly associated with fat and fiber intake. Intervention strate-
gies, targeting beliefs and knowledge may help people adopt essing, and the Health Belief Model. 15 Key elements of these
more healthful diets. The findings on perceived norms suggest theories are that improvements in diet are related to changes
that extrinsic motivations (such as pressure from others) may in (1) beliefs about diet and health, (2) knowledge and skills
be less effectIve than intrinsic mOtIvatIOns (such as beliefs) on related to selecting healthful diets, and (3) social norms
promoting healthful behaviors. regarding expectations about an individual's behavior.
Here, we give a brief overview of these constructs .
(fNE 27:86-92, 1995) The beliifthat a particular health practice will be effective
in reducing a threat is an integral part of the Health Belief
Model. 16 Thus, the Health Belief Model assumes that if
INTRODUCTION people can be persuaded that their current health habits are
potentially harmful, they will be motivated to change those
The National Cancer Institute (NCI) has recommended health habits. Information campaigns assume that lack of
that Americans decrease their consumption of fat to 30% or knowledge is a barrier to improving health behaviors, espe-
less of total calories and increase fiber consumption to 20 cially for a health practice as complex as diet. Knowledge
to 30 grams per day.! These recommendations are based on about diet and cancer can be general, such as knowledge of
dietary recommendations, or specific, such as information
on the fat and fiber composition of foods. Finally, a com-
The research was supported by Grant CA-34847 (Cancer Preventton Research ponent of the Theory of Reasoned Action 17 suggests that
Program) and Grant CA-09661 (Cancer PreventIOn Traming Program) from the
N atIOnal Cancer Instltute. perceived norms (behavioral expectations of peers and family
Address for correspondmg: Ruth E, Patterson, Ph.D, R,D , Cancer Prevention members) have an influence on health behaviors. More
Research Program, Fred Hutchmson Cancer Research Center, 1124 ColumbJa,
MP 702; Seatte, WA 98104; Tel: (206) 667-5014, Fax (206) 667-5977.
detailed descriptions of the theoretical motivations for these
© 1995 SOCIETY FOR NUTRITION EDUCATION constructs are given by Glanz.!8

86
Journal of NutntlOn EducatlOn Volume 27 Number 2 March. Apnl 1995 87

Some researchers have studied how psychosocial con- selected telephone number (i.e., household), one eligible
structs relate to healthful eating patterns,19 low-calorie, adult of each gender was mterviewed. In female- or male-
low-fat eating behavior during a weight loss course,20 and only households, only one person was interviewed. To
health-oriented beverage consumption. 21 Only recently select participants, a modification of the Troldahl and Car-
have researchers addressed predictors of selecting low-fat, ter28 procedure was used in which selection of a random
hIgh fruit and vegetable diets. 22 - 25 In general, there is some adult is based on the total number of adults in the household
consistency in findings that beliefs in a strong relationship and the number of men and/or women. A series of tables
between diet and health and personal motives to eat a are then used to specity the gender and age range of the
healthful diet predict healthful diet behaviors. It is less clear selected respondent (e.g., "oldest male" or "second oldest
whether environmental factors, especially social support or female").
perceived norms, influence dietary habits.
We hypothesized that certain psychosocial constructs Survey instrument. The annual CRBS survey consists
(belief in an association between diet and cancer, knowl- of a core set of questlOns on risk factors for cancer, including
edge of diet recommendatlOns, knowledge of fat and fiber dietary habits, sun exposure, alcohol consumption, smoking
composition of foods, and perceived no=s) are positively behavior, preventive cancer screening, and self-rated health
related to healthful diets. We examined this question using and standard demographic items. The questionnaIre was
data from a random digit dial survey sample, representative administered using a computer-aided telephone interview
of the state of Washington. Population-based samples are (CATI) system and took an average of 25 minutes to
important because much of the early research on improving complete. (Copies of the survey instrument are available
individuals' diets was performed in traditional clinical set- from the authors.) The response rate (completed interviews
tings, with small, select samples and individual counseling divided by known elIgibles) was 0.75 for females and 0.61
techniques; therefore, neither the techniques nor the results for males, and the conservatively estimated effective rate
can be directly applied to community-based nutrition inter- (completed interviews divided by estimated eligibles) was
vention programs. 13 This study can help health promotion 0.65 for females and 0.53 for males. 26 These response rates
practitioners and program planners develop, focus, and are somewhat less than optimal and may raise concerns
target community-based interventions by testing the asso- about the generalizability of study results. Therefore, in a
ciations used in the design of public health interventions. separate but related study, we used enhanced calling efforts
to increase overall survey response rate. We were able to
increase the response rate by 11 %, and observed that there
METHODS were some demographic differences among the participants
reached at different levels of calling effort. However, there
Participants. Study participants were randomly selected were no consistent associations of calling effort with health
adults (18 years of age and older) residing in Washington behavior related to dIet. 26
state. Data were collected from August 1989 through April
1990 ~s part of the Washington State Cancer Risk Behavior Measures. We measured belief in the assooatlOn be-
Survey (CRBS), a random digit dial survey to monitor tween diet and cancer by asking partiCIpants if they believed
cancer risk behavior. The CRBS was part of a NCI -funded that there was an association and, if they responded yes,
program project on the primary prevention of cancer. Goals whether the association was weak, moderate, or strong. For
of the CRBS are to monitor changes in behavior and analysis, we grouped no association with weak.
attitudes related to cancer risk and prevention, with empha- We developed a measure of knowledge of NCI diet
sis on diet, smoking, and screening. Details of sample recommendations using the following procedure. We told
selection and statistical methods have been published. 26 participants that NCI has made recommendations about
how people can change their diet to lower their risk of
Sample frame. Telephone numbers were selected using disease. We asked if they could recall any of these as specific
a modified, two-stage Waksburg method27 in which random food recommendations (yes/no) and, if so, what these
four-digit numbers were affixed to each telephone prefix recommendations were (open-ended response). Partici-
within Washington state's two area codes. The sample of pants were categorized as having no knowledge if they were
telephone numbers was selected to be representative of the unable to recall any diet recommendation. Participants were
entire state and was based on the distribution of residential coded as having some knowledge if they recalled any
lines for each prefix. For each telephone number in the sensible diet recommendation (e.g., "eat more chicken" or
sample, 11 attempts to contact an individual were made in "eat more carrots"). Participants were categorized as having
the first month of calling. Once it was dete=ined that the knowledge of the fat and/ or fiber goal if they stated the goal
phone number reached a household with at least one adult, explicitly ("eat less fat" or "eat more fiber") or mentioned
a screening process was used to select at random one man two or more diet changes relating to a single goal. For
and one woman from each household. For each randomly example, if a participant answered "eat less butter" and "eat
88 Patterson et al./DIET-CANCER BELIEFS AND HEALTHFUL DIETS

less red meat, " they were coded as having knowledge of the 0.43, and 0.47 for dietary fiber. 33 The correlations for fiber
NCI diet recommendation to eat less fat. We did not were less than optimal and may introduce additional noise
require nutritionally precise answers (e.g., less than 30% in our nutrient estimates. Because data collection was in the
kcals from fat) because our purpose was to obtain a measure fall and winter, our point estimates for fat and fiber intake
of general knowledge about foods/nutrients and cancer risk. may be slightly biased because of seasonal reporting bias 34;
We developed a measure of practical knowledge about however, this should not affect the relationship between the
how to select a healthful diet using questions on the fiber dietary variables and psychosocial factors.
and fat content of food. For 12 foods, participants were
asked which foods they thought were high in fat (white Statistical analyses. Descriptive analyses of the distribu-
bread, whole milk, butter, margarine, etc.) For another 12 tion of diet and cancer beliefs, knowledge of NCI diet
foods, participants were asked which foods they thought recommendations, knowledge of food composition, and
were high in fiber (white bread, skim milk, baked potatoes, perceived norms by age and education (Table 1) are
etc.) We summed the number of correct responses and weighted for the complex survey design, and these results
categorized the summary score as low (0-15 correct re- are representative of the Washington state population. As
sponses), intermediate (16-19 correct responses), and high required by the design of the survey, these results are
knowledge (20-24 correct responses). presented separately for males and females. The derivation
In this report, we define perceived norms as a measure of the sample weights IS given in Kristal et aJ.26 We used
of perceived pressure to adopt a healthy diet. This definition chi-square tests to detect the presence of a statistically
is related to the construct of extrinsic motivation as defined significant association of age and education with diet and
by Curry et al., 29 who state that extrinsically motivated cancer beliefs, knowledge, or perceived norms. Statistical
behaviors are those that a person performs to receive some tests in Table 1 were based on weighted data, but the
extrinsic reward, while intrinsically motivated behaviors are weights were standardized so the sum of the weights
ones for which the rewards are internal to the person. We equaled the number of persons interviewed.
measured perceived norms from five questions on whether We used multiple linear regression to model the effects
the participant felt pressure to eat a healthy diet from the of belief, knowledge, and perceived norms on the dietary
people they live with, friends, coworkers, media, and their outcomes. Results related to relationships between beliefs,
doctor. We scored the responses as 3 (a lot), 2 (some), or 1 knowledge, perceived norms and diet (Table 2) were un-
(none) and used the average as a summary measure. This weighted but adjusted for the following covariates: gender,
perceived norms scale had an alpha coefficient of 0.62, an age (three categories: 18-34,35-59,60+), education (years)
indication of good internal consistency.30 and self-rated health (four categories: poor/fair, good, very
Our outcome measures were three measures of diet: good, excellent). We categorized age because of its non-
self-reported healthful diet changes, percent energy from linear association with diet. We also included a gender by
fat, and dietary fiber. We assessed diet changes by asking age interaction in all models, which was the only statistically
participants whether their intake of 12 foods had decreased, significant pairwise interaction among the covariates. A
increased, or stayed the same over the previous 5 years. For linear contrast was used to test for trend across the categories
this measure, we chose a mix offive foods high in fat (whole of belief, knowledge, and perceived norms.
milk, butter, margarine, cheese, eggs, red meat), five foods
high in fiber (fruits, vegetables, whole wheat bread, whole
grains), and two foods that were neither (fish and alcohol). RESULTS
These items were developed by White et al. 31 to examine
dietary change among husbands of participants in a low-fat Study sample characteristics. Of the 1967 survey par-
dietary intervention and have proven useful in other studies ticipants with complete data used in this research, 33% were
of dietary change. 32 We summed the number of healthful aged 18-34, 47% were 35-59, and 21 % were 60 or older.
changes to make a scale from 0 to 12. Forty-two percent of the participants were males, 95% were
We assessed current diet quality using the Quick Dietary white, 59% had some college or post-secondary education,
Screen (QDS). The QDS consists of three sets of items: 16 and 86% rated their own health as good or excellent.
questions about whether a food was eaten daily, 28 ques-
tions about whether a food was eaten at least weekly, and Diet-cancer beliefs, knowledge, and perceived
2 questions about frequency of consumption of red meat norms. The distribution of diet-cancer beliefs, knowl-
and use of butter or margarine in breads. Details of the edge of NCI diet recommendations, knowledge of food
development of this assessment instrument have been pub- composition, and perceived norms by age and education for
lished. 33 In a validation study, correlations between a tele- men and women are given in Table 1. Age was significantly
phone-administered QDS and two 4-day food records, a related to all four constructs. Adults aged 35 to 59 had the
food frequency questionnaire, or the average of both were strongest belief in the association of diet and cancer and the
0.54, 0.65, and 0.65 for percent energy from fat and 0.38, most knowledge about NCI recommendations and food
Journal of Nutrition Education Volume 27 Number 2 March. April 1995 89

Table 1. Distribution (percent) of belief, knowledge, and perceived norms related to diet and cancer by sex, age, and education (adjusted to
Washington state 1989 census).

Females Males

Age (Column Percents)

18-34 35-59 60+ 18-34 35-59 60+

N 360 533 250 282 397 145


Belief in diet-cancer connection
None-weak 34a 24 50 43 a 33 57
Moderate 44 37 24 36 39 22
Strong 22 38 25 21 28 21
Knowledge of National Cancer Institute diet recommendations
None 21 a 12 32 32a 22 44
Some 17 12 17 19 14 15
Fat or fiber goal 41 39 28 31 39 22
Fat and fiber goal 20 36 23 17 25 19
Knowledge of food composition
Low 14a 7 21 14b 22 20
Intermediate 52 49 53 57 46 56
High 34 45 26 28 31 24
Perceived pressure to eat a healthy diet
None 9a 15 37 15a 17 41
Some 28 40 34 43 38 26
Strong 62 45 29 42 45 33

Years of EducatIOn (Column Percents)

:s: 12 13-15 16+ :s: 12 13-15 16+

N 513 327 303 285 242 297


Belief in diet-cancer connection
None-weak 46 a 27 18 52a 47 25
Moderate 29 41 46 26 32 45
Strong 25 32 36 21 21 30
Knowledge of National Cancer Institute diet recommendations
None 28 a 18 6 42 a 30 19
Some 20 12 9 14 20 15
Fat or fiber goal 36 40 37 29 31 38
Fat and fiber goal 16 30 48 14 18 29
Knowledge of food composition
Low 19a 7 6 28 a 19 11
Intermediate 55 52 42 51 59 47
High 26 41 52 21 22 42
Perceived pressure to eat a healthy diet
None 19 15 20 22 c 24 14
Some 34 36 33 39 36 38
Strong 47 49 47 39 40 47

a Chi-square test for distribution statistically significant (p < .001).


bChi-square test for distribution statistically significant (p < .01).
c Chi-square test for distribution statistically significant (p < .05).
90 Patterson et al./DIET-CANCER BELIEFS AND HEALTHFUL DIETS

Table 2. Relations of diet-cancer belief, knowledge, and social norms to healthful diets (adjusted for gender, age, and education and self-related
health).

Diet Measures (Means)

Na Healthful % Energy Fiber


Changes b from Fat (Grams)

Belief in diet-cancer connection


None-weak 645 4.9 35.2 12.9
Moderate 678 5.1 35.1 12.7
Strong 482 6.0 33.5 14.0
P for trend 0.001 0.001 0.001
Knowledge of National Cancer Institute diet recommendations
None 490 4.2 35.6 12.7
Some 299 5.0 35.6 12.6
Fat or fiber goal 689 5.7 34.4 13.2
Fat and fiber goal 489 59 34.0 13.4
P for trend 0.001 0.001 0.001
Knowledge of food composition
Poor 274 4.0 36.3 12.4
Fair 1036 5.4 34.9 13.1
Excellent 657 5.6 342 13.3
P for trend 0.001 0.001 0.001
Perceived pressure to eat a healthy diet
None 361 4.1 35.0 13.1
Some 719 5.0 35.1 12.8
A lot 887 5.9 34.6 13.2
P for trend 0.001 0.06 0.004

a Sample sizes vary due to missing data.


b Number of changes to lower fat or higher fiber foods that people reported making over the past 5 years.

composition, while older adults (60+) had the lowest levels among participants with college degrees, only 18% of fe-
of belief, knowledge, and perceived norms. For example, males and 25% of males believed that there was no connec-
50% of older females and 57% of older males believed that tion between diet and cancer, 6% and 19% know no NCI
there was no connection between diet and cancer compared diet recommendations, and 6% and 11 % had low knowl-
to only 24% and 33% (females and males, respectively) of edge of food composition. Education had little effect on
middle-aged adults and 34% and 43% of younger adults. perceived nonns, with approximately 50% of females and
Young women (18-34) perceived the greatest pressure to 40% of males at all education levels reporting feeling strong
eat a healthy diet, with 62% reporting that they perceived pressure to eat a healthy diet.
a lot of pressure to eat a healthy diet. Conversely, among
older adults, only about 30% reported feeling strong pres- Associations between diet-cancer beliefs, knowl-
sure to eat a healthy diet. edge, perceived norms, and diet. The relationships
Diet-cancer beliefs and knowledge were positively asso- of diet-cancer beliefs, knowledge of NCI recommenda-
ciated with higher education (p < .001). Perceived nonns tions, knowledge of food composition, and perceived
were not significantly associated with education in females norms with three measures of diet (controlled for gender,
and only marginally for males. Among participants with age, education, and self-rated health) are given in Table 2.
high school or less education, about 50% of men and After adjustment for sociodemographic covariates, all four
women believed that there was no connection between diet constructs were significantly associated with reported
and cancer, 28% of females and 42% of males had no healthful diet changes. The number of healthful diet
knowledge of NCI diet recommendations, and 19% and changes reported by participants appeared to be most influ-
28% had low knowledge of food composition. However, enced by perceived norms. Individuals who perceived
Journal of Nutntion EducatIOn Volume 27 Number 2 March. April 1995 91

strong pressure to eat a healthful diet made 1.8 more Few studies have examined the relationships between
healthful changes (5.9 vs. 4.1) in the past 5 years compared diet-cancer beliefs, knowledge, and perceived norms with
to those who felt no pressure (p for trend:::; .001). Among healthful diet, and we are not aware of any published
the predictors of dietary intake, the largest improvements research using a representative, population-based sample.
(decreases in fat intake and increases in fiber intake) were Our results indicate that beliefs and knowledge are signifi-
among individuals with good knowledge of food composi- cantly associated with having a healthful diet. Comparisons
tion. The largest improvements in diet were seen for beliefs of our results with other studies are difficult because of
and knowledge. Participants with strong beliefs in a diet- differences in sample and measures, In a previous study, we
cancer connection consumed 1.1 grams more fiber than used carefully designed scales in a select sample of post-
those with no belief (p for trend:::; .001), and participants menopausal women and found that the psychosocial factors
with excellent knowledge of food composition consumed most strongly associated with selecting low-fat diets were
2.1 % less energy from fat than those wIth low knowledge related to knowledge about fat in foods and perceived
(p for trend :::; .001). Although perceived norms were norms,24 Conversely, a study of a convenience sample of
strongly associated with reported healthful diet changes, 210 English adults found that knowledge offood composi-
there was only a weak association with eStimated fat and tion was not associated with frequency of intake of five high-
fiber intake. fat foods, and personal attitudes were much more important
than perceived norms in predicting consumption of high-fat
foods. 39 Glanz et aJ.25 examined similar psychosocial items
DISCUSSION in a pilot study of 652 working men and women and found
that beliefs, food composition knowledge, and social sup-
Results of this random digit dial survey suggest that most port were weakly associated with fat or fiber in the diet,
Washington state residents believe that what they eat is Perceived norms were strongly associated with reported
related to their chance of getting cancer. Further, we found healthful diet changes, but they were only weakly associated
that state residents were fairly knowledgeable about the with dietary fat and fiber intake, It may be that questions
NCI diet recommendations and fat and fiber composItion about healthful diet changes are subject to considerable
of food and felt pressure from their family, friends, and social desirability bias, The negative findings on perceived
coworkers to eat a healthy diet. These findings are similar norms and healthful diet appear to echo the work in smok-
to the Cancer Control Supplement of the 1987 National ing cessation by Curry et aL;o suggesting that extrinsic
Health Interview Survey (NHIS), in which 73% of Ameri- motivations (such as reward or perceived pressure by others)
cans agreed that cancer may be related to what people eat may be less effective than intrinsic motivations (such as
or drink. 35 Participants in the NHIS were also asked open- beliefs) on behavior change, She found that the rate of
ended questions on foods that they thought affected a continuous smoking abstinence at 1-year post-intervention
person's risk of getting cancer. These results were very was more than twice as high in the group receiving intrinsic
similar to our findings in that individuals who were most versus extnnsic motivation strategies,
likely" to correctly identify foods to eat more (or less) to One caveat to this study concerns the QDS. This instru-
prevent cancer were females, adults aged 35 to 49 (as ment is a relatively brief assessment tool; therefore, the
compared to younger and older adults), and more educated associatlOns between dietary intake and psychosocial meas-
participants. ures were not as strong as would be expected had we used
It is notable that people aged 60 and older were least longer, more intensive instruments such as food records or
likely to believe in a connection between diet and cancer, frequencies.
have knowledge of NCI diet recommendations or food Young people, older adults, males, and the less educated
composition, or to feel pressure to eat a healthy diet. This appear to be important targets for nutrition interventions,
finding is in contrast to research that indicates that older Our results suggest that strategies that focus on beliefs about
people are very interested in their health and willing to diet and health and knowledge about fat and fiber content
improve their behaviors to improve their health. 36 ,37 In of food may help people adopt more healthful diets. How-
focus group discussions with older Americans performed by ever, it is important to note that this is a cross-sectional
the Public Health Service, nutrition was an issue of wide- study, and therefore it is not possible to determine true
spread concern; however, most individuals were unable to causality, A belief in a diet and disease connection and a
describe what constitutes a balanced diet and there was healthy diet may both result from underlying factors that
confusion over basic issues such as what fiber is or whether were not measured in this study. In addition, much addi-
red meat should be included in one's diet,38 Additional tional research is needed on strategies for affecting beliefs
research should be done to indicate whether community- and knowledge in a community, Results from ongoing
based interventions need to be specifically focused and trials, such as the Working Well Study,!! should further
tailored to have a positive impact on the diet of elderly elucidate the optimal components and strategies for com-
Americans. munity-based diet change interventions.
92 Patterson et al.lDIET-CANCER BELIEFS AND HEALTHFUL DIETS

REFERENCES 21 Lewis C], S,ms LS, Shannon B. Exammation of specific nutrI-


tIOn/health behaVIOrs usmg a SOCIal cogmtive model.] Am DIet Assoc
1. Butrum RR, Clifford CK, Lanza E. NCI dIetary gUIdelines: rationale. 1989; 89:194-202.
Am] Clm Nutr 1988; 48:888-95. 22. Brlnberg D, Axelson ML. Increasmg the consumptIOn of dIetary fiber.
2. Stemmetz KA, Potter ]D. Vegetables, frUIt, and cancer I. EpIdemIOl- a declSlon theory analYSIS. Health Educ Res 1990; 5:409-20.
ogy. Cancer Causes Control 1991; 2:325-57. 23. Trenkner LL, Rooney B, V,swanath K, et al. Development of a scale
3. National Research CounCIl, CommIttee on D,et and Health Diet and using nutritIOn attItudes for audIence segmentation. Health Educ Res
health. ImplicatIOns for reducmg chromc d,sease. Washmgton, DC: 1990; 5'479-87.
NatIOnal Academy Press, 1989 24. KrIStal AR, Bowen D], Curry S], Shattuck AL, Henry HJ. Nutrition
4. WIllett WC, MacMahon B. D,et and cancer-an overvIew (part 1). knowledge, attItudes and perceIved norms as correlates of selectmg
N Engl] Med 1984; 310:633-738. low-fat diets. Health Educ Q 1990; 5:467-77.
5. Willett WC, MacMahon B. D,et and cancer-an overVIew (part 2). 25. Glanz K, KrIStal AR, Sorensen G, Palombo R, Helmendinger ],
N Engl] Med 1984; 310:697-703. Pro bart C. Development and validation of measures of psychosocial
6. Goldberg ]P. NutritIOn and health commumcatIOn: the message and factors mfluencmg fat- and fiber-related d,etary behavIOr. Prev Med
the medIa over half a century. Nutr Rev 1992; 50:71-7. 1993; 22373-87.
7. Ernst ND, Wu M, Frommer P, et al. Nutrition education at the pomt 26. Kristal AR, Wh,te E, DaVIS ]R, et al. The effects of enhanced calling
of purchase: the Foods for Health Project evaluated. Prev Med 1986; efforts m random digIt dIal surveys m response rates and populatIon
15:60-73. and populatIOn level estImates of health behaVIOr and costs. Public
8. Kottke T, Puska P, Salonen ]T, et a1. Changes in perceIved heart Health Rep 1993; 108'372-9.
d,sease risk and health dUring a commumty-based heart d,sease pre- 27. Waksburg J. SamplIng methods of random dIgIt dialIng. ] Pers Soc
ventIOn program: the North KarelIa project. Am] PublIc Health 1984; Psychol 1978, 39:806-20.
74.1404-5. 28 Troldahl VC, Carter RE. Random selectIOn of respondents WIthin
9. Wagner EH, Koepsell T, Anderman C, et a1. The evaluation of the households in phone surveys.] Market Res 1964; 1:71-6.
Kaiser FamIly FoundatIOn Health PromotIOn Grants Program: overall 29. Curry S, Wagner EH, Grothaus LC. Intnnsic and extrinsIC motivation
desIgn.] ClIn EpldemIOl1991; 44:685-99. for smokmg cessatIOn.] Consult Clm Psychol 1990; 58:310-6.
10 Hebert ]R, Harris DR, Sorensen G, et al. A work-SIte nutritIon 30 Nunnally]C. Psychometric theory 2nd ed. New York: McGraw-
interventIOn: Its effects on the consumptIOn of cancer-related nutn- HIll, 1978
ents. Am] Public Health 1993; 83'391-4. 31. WhIte E, Henderson MM, HurlIch M, et al. Dietary changes among
11. Abrams D, Boutwell B, Gnzzle ], et a1. Cancer control at the husbands of participants m a low-fat chetary mterventIOn. Am] Prev
workplace: The Workmg Well Trial. Prev Med 1994; 23.15-27. Med. 1991; 7:319-25
12. Havas S, Helmendinger ], Reynolds K, et al. 5 A Day for Better 32. Shattuck AL, Wh,te E, Knstal AR How women's adopted low-fat
Health: a new research Imtlatlve.] Am D,et Assoc 1994; 94:32-9. diets affect theIr husbands. Am] PublIc Health 1992, 82:1244-50.
13. Kristal AR. NutrItIOnal mterventIOn. How can we meet the "Healthy 33. Knstal AR, Shattuck AL, Henry H], Fowler AS. Rapid assessment of
People 2000" goals? Med Exerc Nutr Health 1993; 2:3-4 dIetary mtake of fat, fiber, and saturated fat: valIchty of an mstrument
14. Gl~nz K. NutritIOn for nsk factor reductIOn and patIent educatIOn: a SUItable for community mtervention research and nutntIOnal surveil-
review. Prev Med 1985; 14:721-52 lance. Am] Health PromotIOn 1990; 4.288-95.
15. Glanz K, Enksen MP. IndIVIdual and commumty models for chetary 34. Subar AF, Frey CM, Harlan LC, Kahle L DIfferences m reported food
behaVIOr change] Nutr Educ 1993; 25:80-6. frequency by season of questIOnnaire administration' The 1987 Na-
16. Rosenstock 1M The Health Belief Model: Explammg health behavior tional Health Interview Survey. EpIdemIOlogy 1994; 5:226-33.
through expectancies. In: Glanz K, LeWIS FM, R,mer BK, eds., Health 35. Cotugna N, Subar A, Helmendmger], Kahle L. Nutntion and cancer
behaVIOr and health education: theory, research, and practIce. San preventIOn knowledge, beliefs, attItudes, and practIces: The 1987
FranCISCO: ]ossey-Bass, 1990:39-60. NatIOnal Health Interview Survey.] Am Diet Assoc 1992; 92:963-8
17. AJzen I, Flshbem M. Understanding attitudes and predicting social 36. Heckler MM. Health promotIOn for older Americans. Public Health
behavIOr. Englewood Cliffs, NJ PrentIce-Hall, 1980 Rep 1985; 100:225-30.
18. Glanz K. The dietary fat message and the process of behavior change. 37. Goodwm ]S, Leonard AG, Hooper EM, Garry P] Concern about
In: The PublIc Health SerVIce report on d,etary fat and health. cholesterol and ItS aSSOCIatIon WIth d,et m a group of healthy elderly.
Washmgton, DC: Office of D,sease Prevention and Health Promo- Nutr Res 1985; 5:141-8.
tIOn, PHS, 1994 (m press). 38. Maloney SK. Aging and health promotIOn: market research for publIc
19. Contento IR, Murphy BM. Psycho-social factors dIfferentIating peo- educatIOn. ExecutIve summary. Washmgton, DC: PublIc Health Serv-
ple who reported making deSIrable changes m theIr d,ets from those ice, 1984.
who dId not.] Nutr Educ 1990; 22:6-14. 39. Shepherd R, Stockley L. NutritIOn knowledge, attItudes, and fat
20. Shannon B, Bagby R, Wang MW, Trenkner LL. Self-efficacy- a consumptIOn.] Am Diet Assoc 1987; 87:615-9.
contnbutor to the explanatIOn of eating behaVIOr. Health Educ Res 40. Curry S], Wagner EH, Grothaus LC. EvaluatIOn of mtrmsic and
1990; 5:395-407. extnnslC motivatIOn mterventIOns WIth a self-help smoking cessation
program.] Consult Clm Psychol1991; 59.318-24

You might also like